For image retrieval, it is necessary to add keywords into images for the reference. This paper suggests a new method, with which keywords are automatically added to landscape images. Keywords to be added can be divided into two categories: Blocked keyword and Global keyword. Blocked keyword is added to each square block that is created by dividing an image. The keywords include the colors, edges, and position of each block. On the other hand, Global keyword is obtained from the Blocked keywords added already. They include the area, the circular rate, and the gravity center position of the regions, which have the maximum area for each blocked keyword added in the image. Using those features, keywords are added by discriminant analysis. The experiment of the method is performed and the result is reported in this paper.
View full abstract